Utilizing OSDL's STP for Reiser4 testing/benchmarks?

Utilizing OSDL's STP for Reiser4 testing/benchmarks?

[Mike Benoit]

[-- Attachment #1: Type: text/plain, Size: 1014 bytes --]

Has the Reiser4 team looked at utilizing the OSDL Scalable Test Platform
(STP) service to benchmark and test Reiser4 patches? They seem to offer
a wide variety of hardware to test on, and already have some file system
benchmarks available to choose from. 

I noticed the Namesys benchmark page hasn't been updated in quite some
time, understandably so, as benchmarking can be incredibly time
consuming.  Perhaps using the STP would make this much easier to do more
frequently?

It would be really nice to see how the batch_write/read(?) patches to
the kernel affect things, and how Reiser4 performance progresses over
time. I know Hans has mentioned CPU usage dropping in recent patches,
but actual numbers from a benchmark are usually more compelling to see.

User Guide:
http://www.osdl.org/lab_activities/kernel_testing/stp/guide.html/document_view

Test Details:
http://www.osdl.org/lab_activities/kernel_testing/stp/test_details.html/document_view

-- 
Mike Benoit <ipso@snappymail.ca>

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 189 bytes --]

Re: Utilizing OSDL's STP for Reiser4 testing/benchmarks?

[Hans Reiser]

Elena, on Monday can you comment on this in detail?

Thanks,

Hans

Mike Benoit wrote:

>Has the Reiser4 team looked at utilizing the OSDL Scalable Test Platform
>(STP) service to benchmark and test Reiser4 patches? They seem to offer
>a wide variety of hardware to test on, and already have some file system
>benchmarks available to choose from. 
>
>I noticed the Namesys benchmark page hasn't been updated in quite some
>time, understandably so, as benchmarking can be incredibly time
>consuming.  Perhaps using the STP would make this much easier to do more
>frequently?
>
>It would be really nice to see how the batch_write/read(?) patches to
>the kernel affect things, and how Reiser4 performance progresses over
>time. I know Hans has mentioned CPU usage dropping in recent patches,
>but actual numbers from a benchmark are usually more compelling to see.
>
>User Guide:
>http://www.osdl.org/lab_activities/kernel_testing/stp/guide.html/document_view
>
>Test Details:
>http://www.osdl.org/lab_activities/kernel_testing/stp/test_details.html/document_view
>
>  
>

Re: Utilizing OSDL's STP for Reiser4 testing/benchmarks?

[Nate Diller]

On 7/21/06, Mike Benoit <ipso@snappymail.ca> wrote:
> Has the Reiser4 team looked at utilizing the OSDL Scalable Test Platform
> (STP) service to benchmark and test Reiser4 patches? They seem to offer
> a wide variety of hardware to test on, and already have some file system
> benchmarks available to choose from.
>
> I noticed the Namesys benchmark page hasn't been updated in quite some
> time, understandably so, as benchmarking can be incredibly time
> consuming.  Perhaps using the STP would make this much easier to do more
> frequently?
>
> It would be really nice to see how the batch_write/read(?) patches to
> the kernel affect things, and how Reiser4 performance progresses over
> time. I know Hans has mentioned CPU usage dropping in recent patches,
> but actual numbers from a benchmark are usually more compelling to see.

the dev team has seen a number of benchmarks which Hans' statements
are based on... maybe we should post some of them publicly?

NATE

Re: Utilizing OSDL's STP for Reiser4 testing/benchmarks?

[Hans Reiser]

Nate Diller wrote:

> On 7/21/06, Mike Benoit <ipso@snappymail.ca> wrote:
>
>> Has the Reiser4 team looked at utilizing the OSDL Scalable Test Platform
>> (STP) service to benchmark and test Reiser4 patches? They seem to offer
>> a wide variety of hardware to test on, and already have some file system
>> benchmarks available to choose from.
>>
>> I noticed the Namesys benchmark page hasn't been updated in quite some
>> time, understandably so, as benchmarking can be incredibly time
>> consuming.  Perhaps using the STP would make this much easier to do more
>> frequently?
>>
>> It would be really nice to see how the batch_write/read(?) patches to
>> the kernel affect things, and how Reiser4 performance progresses over
>> time. I know Hans has mentioned CPU usage dropping in recent patches,
>> but actual numbers from a benchmark are usually more compelling to see.
>
>
> the dev team has seen a number of benchmarks which Hans' statements
> are based on... maybe we should post some of them publicly?
>
> NATE
>
>
Sure.  I asked for that, actually.